search

weecology / MATSS-forecasting

Forecasting Analysis Comparison for Ecological Time Series
https://weecology.github.io/MATSS-forecasting/
Other
5 stars 1 forks source link

Scientific Questions #10

Open ha0ye opened 5 years ago

ha0ye commented 5 years ago

  1. How to select an ideal forecasting method? (~ Ward et al. 2014)

    • evaluate forecasting approaches across different time series
    • do time series properties influence which methods (or parameter settings) perform well?
      • measures of linearity vs. nonlinearity

  2. meta-analysis of time series properties and relationship with permutation entropy and forecasting performance

    • similar to 1, but maybe more of a macroecological focus on life history traits, geographic location, environmental covariates?

  3. Ensemble forecasting methods

    • automated vs. human-informed parameter selection
    • how to best select embedding dimension parameters for forecasting

  4. Effective forecast horizon

    • changes in predictability over different forecast horizons
    • relationship with data properties, etc.

  5. Transfer of models across populations

    • we have lots of data from the same species - how well do models transfer from one population to another?
      • also, various kinds of hierarchical or multi-population models
opetchey commented 5 years ago

Sorry to have not previously engaged...

Questions 2 and 5 seem particularly interesting and at first sight reasonable feasible... model transferability is a obviously super important to know, and one could couple this with if differences/similarities in geography&environment explain transferability.

ha0ye commented 5 years ago

Sorry to have not previously engaged...

Questions 2 and 5 seem particularly interesting and at first sight reasonable feasible... model transferability is a obviously super important to know, and one could couple this with if differences/similarities in geography&environment explain transferability.

No problem! We're still in a bit of setting up phase.

Are you going to be at the Eco-forecasting meeting in Washington DC next week?

opetchey commented 5 years ago

Unfortunately not... its semester here.